To test your graphics card for errors, download FurMark 2 and run its GPU stress test for 20 to 30 minutes while monitoring temperature, frame rate, and visual artifacts. If the test completes without crashes, artifacts, or overheating above 90°C, your GPU is functioning properly. If any of those occur, the card likely has a hardware issue.
Applies to: Windows 10 (22H2) and Windows 11 (23H2, 24H2, 25H2) | Last updated: April 9, 2026
Key Takeaways
- FurMark 2 is the go-to free GPU stress test tool — it pushes your graphics card to 96%+ utilization and reveals instability that normal use would not trigger
- Run the stress test for 20-30 minutes — a healthy GPU should maintain stable temperatures (typically 60-85°C) without visual artifacts, crashes, or system shutdowns
- Artifacts (weird lines, colors, or flickering) during the test mean the GPU is failing — this is a hardware issue that cannot be fixed with software
- Check your power supply first — an underpowered PSU can mimic GPU failure symptoms, so verify your PSU meets the GPU manufacturer’s recommended wattage before testing
- Confirm the GPU is detected in Device Manager before stress testing — if the card does not show under Display Adapters, the issue is a driver or connection problem, not a stress test scenario
Quick Steps
- Open Device Manager (right-click Start > Device Manager) and confirm your GPU appears under Display Adapters with status “This device is working properly”
- Verify your power supply wattage meets or exceeds your GPU manufacturer’s recommendation
- Download FurMark 2 from the official website, extract the zip, and launch FurMark_GUI.exe
- Select your monitor resolution and the GPU you want to test, then click Run
- Let the test run for 20-30 minutes — watch for artifacts, crashes, shutdowns, or temperatures above 90°C
- Press Escape to stop the test and review results — if errors were detected, the GPU likely needs replacing
How to Check if Your GPU Is Detected in Device Manager
Before running any stress test, you need to confirm that Windows actually sees your graphics card. Right-click the Start button and select Device Manager. Expand the Display adapters section — your GPU should be listed there by name (for example, “NVIDIA GeForce RTX 3060” or “AMD Radeon RX 7600”).
Double-click on the GPU and check the Device status field. It should say “This device is working properly.” If it shows an error code, or if your GPU is not listed at all, you have a driver or hardware connection problem that needs to be resolved before stress testing makes sense.
If the driver is missing or shows an error, follow my guide on how to install or update your graphics driver to get the correct driver installed. Once the GPU shows up properly in Device Manager, you can proceed with the stress test.
How to Check if Your Power Supply Is Sufficient
An underpowered power supply (PSU) can cause your graphics card to crash under load, produce artifacts, or shut down your entire system — symptoms that look identical to a failing GPU. Before blaming the graphics card, you need to rule out the PSU as the cause.
Every GPU has a recommended minimum PSU wattage listed on the manufacturer’s website. For example, an NVIDIA RTX 3060 recommends a 550W power supply. You always want to have a bit more than the minimum — if your GPU recommends 550W, a 650W PSU gives you comfortable headroom.
Warning: If your power supply does not meet the recommended wattage for your GPU, do not run a stress test. A stress test pushes the card to maximum power draw, and an insufficient PSU under that load can damage components or cause sudden shutdowns. Upgrade your PSU first.
To check your PSU wattage, look at the label on the side of the power supply unit inside your PC case. It will list the total wattage (e.g., 650W). If you are not sure what PSU you have, check your purchase receipt or system specifications from the manufacturer.
How to Download and Set Up FurMark 2
FurMark is a free GPU stress testing tool that has been the industry standard for years. It renders an extremely demanding 3D scene (a furry donut) that pushes your GPU to its absolute limit — far beyond what any game or application would demand. This makes it ideal for exposing hardware faults.
Go to the official FurMark website and download FurMark 2 (the latest version). Grab the 64-bit zip file rather than the installer — it is portable and does not require installation. Extract the zip to any folder and look for FurMark_GUI.exe inside.
https://www.geeks3d.com/furmark/Launch FurMark_GUI.exe. The interface is straightforward — you will see options for resolution, GPU selection, and the test controls. No additional configuration files or dependencies are needed.
How to Configure and Run the GPU Stress Test
In FurMark’s main window, set the resolution to match your monitor (for most people this is 1920×1080). If you have multiple GPUs (for example, integrated graphics plus a dedicated card), use the GPU dropdown to select the specific graphics card you want to test.
Click Run to start the stress test. FurMark will open a fullscreen window rendering the fur donut scene while displaying real-time statistics. Let the test run for 20 to 30 minutes minimum. This duration is important — some GPU faults only appear after the card has been under sustained load and the temperature has fully stabilized.
While the test runs, watch for these four failure indicators:
- Visual artifacts — weird colored lines, flickering pixels, blocky patches, or distorted geometry on screen
- Test crashes or closes — FurMark suddenly stops or the display driver crashes and recovers
- Computer shuts down or restarts — the system powers off completely under load
- Temperature exceeds 90°C and keeps climbing — indicates a cooling problem or thermal paste failure
Tip: If you see any artifacts at all — even minor ones — that is a hardware failure. Artifacts during a stress test do not go away with driver updates or software fixes. The GPU silicon itself is damaged.
How to Read Your Stress Test Results
FurMark displays a real-time overlay during the test with all the metrics you need to evaluate your GPU’s health. Here is what each number means and what to look for.
Temperature: This is the most important metric. A healthy GPU under full stress test load typically runs between 60°C and 85°C depending on the card model and cooling solution. Temperatures in the 60s are excellent. Anything below 85°C is perfectly safe. If you are seeing 85-90°C, the card is running hot but still within most manufacturers’ thermal limits. Above 90°C is cause for concern — the card may have a cooling problem, dried-out thermal paste, or blocked fans.
GPU Usage: During the stress test, this should be at 96-100%. If GPU usage is significantly lower, FurMark may not be testing the correct GPU — double-check your GPU selection in the settings.
FPS (Frames Per Second): The exact FPS number does not matter much for a health test. What matters is that it stays stable. If the FPS is steady throughout the test, the GPU is handling the load consistently. Sudden FPS drops or fluctuations can indicate throttling or instability.
VRAM Usage, Power Draw, and Fan Speed: These are useful secondary indicators. Fan speed should ramp up as temperature increases — if it does not, the fans may be failing. Power draw should match your GPU’s expected TDP range. VRAM usage shows how much video memory the test is consuming.
Press Escape at any time to stop the test. If the test ran for 20-30 minutes with stable temperatures, no artifacts, and no crashes — your GPU passed. It is working properly.
When to Replace Your Graphics Card
If FurMark produced artifacts, crashed repeatedly, or caused your system to shut down, and you have already confirmed that your power supply is sufficient, the GPU is almost certainly failing. Unfortunately, there is no software fix for a graphics card with damaged silicon — artifacts during a stress test are a definitive hardware failure.
Before giving up entirely, try these last steps:
- Reinstall the graphics driver — do a clean driver installation using DDU (Display Driver Uninstaller) and then install the latest driver from NVIDIA or AMD. Follow my graphics driver installation guide for the full process.
- Reseat the card — power off the PC, remove the GPU from the PCIe slot, clean the gold contacts gently, and reinstall it firmly. A loose connection can cause similar symptoms.
- Test in another system — if possible, install the GPU in a different computer to confirm the problem follows the card, not the motherboard or PSU.
If the GPU still fails after a clean driver install and reseating, it is time for a replacement. Check out my best graphics cards for every budget guide to find a solid replacement at your price point.
For a more comprehensive system health check beyond just the GPU, AIDA64’s system stability test can stress test your CPU, RAM, and storage simultaneously. And if you are experiencing high temperatures across the board, my guide on checking CPU temperature covers monitoring tools and safe thermal ranges for your processor.
Frequently Asked Questions
Will FurMark damage my graphics card?
No. FurMark pushes the GPU to its maximum rated load, but it does not exceed what the hardware is designed to handle. Modern GPUs have built-in thermal throttling that automatically reduces performance if the temperature gets too high. If FurMark causes a crash or artifacts, the GPU already had a pre-existing fault — the stress test just revealed it.
What temperature is too hot for a GPU during a stress test?
Most GPUs are designed to operate safely up to 83-90°C under heavy load, depending on the manufacturer and model. Temperatures in the 60-85°C range during a FurMark stress test are considered normal. If your GPU consistently exceeds 90°C and keeps climbing, there is likely a cooling issue — check for dust buildup, verify fans are spinning, or consider replacing the thermal paste.
How long should I run the GPU stress test?
Run FurMark for a minimum of 20 to 30 minutes. Some GPU faults only manifest after the card reaches its peak temperature under sustained load, which can take 10-15 minutes. Running for less than 20 minutes may not give the card enough time to expose intermittent issues. If you want to be thorough, running for a full hour is even better.
My computer shut down during the stress test — is the GPU broken?
Not necessarily. A sudden shutdown during a GPU stress test is often caused by an insufficient power supply rather than a faulty graphics card. The stress test pushes the GPU to maximum power draw, and if your PSU cannot deliver enough watts, it will trigger a safety shutdown. Verify your PSU meets the manufacturer’s recommended wattage for your GPU before assuming the card is defective.
Can I use FurMark to test a laptop GPU?
Yes, FurMark works on laptop GPUs. However, laptops have more limited cooling than desktops, so temperatures will run higher and the GPU may thermal throttle sooner. Make sure the laptop is on a flat, hard surface with good airflow. If you have a laptop with both integrated and dedicated graphics, select the dedicated GPU from the FurMark dropdown before running the test.
